CN1136745C - Method for implementing multi-language coding-decoding in universal mobile communication system - Google Patents

Abstract

The invention discloses a method for realizing multiple speech codecs based on a codec algorithm in a universal mobile telecommunication system. The method includes firstly selecting a rate mode according to the code type when coding, and calling a unified AMR speech codec module. If If it is detected as non-silence, it will output speech coding bits according to the determined rate. If it is detected as silence, it will select the respective comfortable noise generation and intermittent transmission modules according to the coding type, and output the coding bits of silence; decoding is the reverse process of coding, only adding A frame error compensation and mute processing steps, and finally output 13bit linear PCM data uniformly. Under the premise of ensuring the same voice quality, based on a voice codec algorithm, the 7 kinds of voice codecs defined by all UMTS system narrowband voice services are uniformly implemented, so as to significantly improve the economy, realizability and reliability of UMTS system voice codecs. Solve the problem of taking up a lot of storage space.

Description

Method for Realizing Multiple Speech Codecs in Universal Mobile Telecommunication System

technical field

The invention relates to a voice coding and decoding technology, and more particularly relates to a method for uniformly realizing multiple voice coding and decoding in a universal mobile telecommunication system (UMTS).

Background technique

The voice codec TC (Transcoder) is the basis for the UMTS system to realize the voice bearer service. It is not only indispensable on the user equipment (UE), but also a functional entity that must be used in the voice bearer equipment in the core network at the present stage. The 3GPP standard organization defines 7 voice codecs for UMTS system to realize narrowband voice services: GSM system enhanced full rate codec (GSM EFR), TDMA system enhanced full rate codec (TDMA EFR), PDC system enhanced full rate codec Codec (PDC EFR), Half Rate AMR Voice Codec (HR AMR), Full Rate AMR Voice Codec (FR AMR), UMTS System AMR Voice Codec (UMTS AMR), UMTS System AMR Voice Codec Standard 2 (UMTS AMR 2) (see 3GPP TS 26.103). Among them, GSM EFR, TDMA EFR, PDC EFR, HR AMR, and F AMR are all voice codec algorithms of traditional 2G systems, and UMTS AMR and UMTS AMR 2 are newly defined voice codecs for UMTS systems. In order to realize intercommunication between the voice bearer equipment on the UMTS core network and other mobile communication systems (GSM, IS54, IS136, PDC, R99 UMTS systems, etc.) in TFO mode, its voice codec must support all the above seven voice codec protocols . UMTS UE also needs to support as many voice codec protocols as possible in order to realize codec non-tandem (Transcoder Free Operation-TFO) mode communication and roaming among multiple systems.

In order to achieve the above-mentioned purpose, the common practice is to implement all seven speech codec algorithms in the speech codec respectively. However, because the speech codec algorithm is very complex, the cost of this implementation method is very expensive (self-developed or purchased). Moreover, the memory space of high-speed semiconductor devices (such as DSP) that is usually used to implement speech codec is very limited, especially UE, which is limited by factors such as cost, power consumption, and volume, and its memory space is even tighter. In order to realize the required program space and data space of 7 kinds of speech codec algorithms, it is considerable and unbearable.

Contents of the invention

The purpose of the present invention is to overcome the defects of the above-mentioned existence, and under the premise of ensuring that the voice quality remains unchanged, based on a voice codec algorithm, the 7 kinds of voice codecs defined by all UMTS system narrowband voice services are uniformly realized, so as to significantly improve the UMTS system. The speech codec is economical and realizable and solves the problem of occupying a large storage space.

The method of the present invention is realized by following technical scheme, realizes the method for multiple speech codecs in the universal mobile telecommunication system, and this method comprises the steps:

When encoding, first judge the current encoding type according to the set command;

Select the rate mode according to the encoding type;

Then call the unified AMR speech encoding module to perform the encoding operation;

Wherein, if it is determined to be silent through the silence detection, then call the respective comfort noise coding (SID) and discontinuous transmission (DTX) modules according to the corresponding coding type;

If it is non-silent, then output speech coding bits according to the determined rate;

Decoding is the reverse process of encoding. First, judge the bad frame indicator (BFI) flag from the received frame, check whether there is an error frame, and judge whether error frame compensation is required according to the bad frame indicator (BFI) flag;

If it is judged to be a wrong frame, perform wrong frame compensation and mute processing;

If the judgment is normal, judge whether it is mute according to the Comfort Noise Code (SID) sign, and if it is mute, call the respective discontinuous transmission and comfort noise decoding modules according to the corresponding codec type;

If it is non-mute, set the rate according to the corresponding codec type, perform AMR decoding operation, and finally output 13bit linear PCM data uniformly.

The determined rate and the set rate according to the corresponding codec type are: GSMEFR selects AMR_12.20, TDMA EFR selects AMR_7.40, PDC EFR selects AMR_6.70, and HR AMR, FRAMR, UMTS AMR and UMTS AMR 2 determines the rate according to the respective rate control command.

Based on the AMR voice codec, the present invention uniformly realizes all seven voice codec algorithms stipulated in the narrowband voice service of the UMTS system, significantly reduces the implementation cost of the UMTS voice codec, solves the problem of storage space, and significantly improves the system implementation efficiency .

Description of drawings

The method of the present invention is further described below in conjunction with accompanying drawing and embodiment:

Fig. 1 is a schematic diagram of the encoding process used by the method of the present invention;

Fig. 2 is a schematic diagram of the decoding process used by the method of the present invention;

Fig. 3 is a statistical table diagram of the amount of code of the traditional implementation method;

Fig. 4 is the code amount statistical tabular diagram of the implementation method of the present invention;

Fig. 5 is a table diagram of storage space comparison of different methods.

Detailed ways

The method for realizing multiple speech codecs in the universal mobile telecommunication system described in the present invention is realized by the following technical solutions. Among the above-mentioned 7 kinds of speech codec algorithms, HR AMR, FRAMR, UMTS AMR and UMTS AMR 2 The core of the algorithm is based on the adaptive multi-rate algorithm, and its principle and structure are the same, but the rate control range of the algorithm is different because of the different application environments. Moreover, the agreement clearly states that UMTS AMR 2 is fully compatible with HR AMR, FRAMR, and UMTS AMR. Therefore, the UMTS AMR 2 protocol can be used to uniformly implement HSAMR, FRAMR, and UMTS AMR.

The AMR algorithm is a multi-rate speech codec algorithm, with a total of 8 modes (AMR_12.20, AMR_10.20, AMR_7.95, AMR_7.40, AMR_6.70, AMR_5.90, AMR_5.15, AMR_4.75), Corresponding to 8 kinds of rates (12.20kbit/s (GSM EFR), 0.20kbit/s, 7.95kbit/s, 7.40kbit/s (TDMA EFR), 6.70kbit/s (PDC EFR), 5.90kbit/s, 5.15kbit /s, 4.75 kbit/s).

In addition, AMR also supports silence detection (VAD) and discontinuous transmission function (DTX). During silence, only comfort noise (carried by SID frame) is transmitted periodically, and no data is transmitted at other times. Therefore, in addition to the above-mentioned 8 speed modes, there are two modes: AMR_SID (1.80kbit/s) and AMR_NODATA (0kbit/s).

GSM EFR, TDMA EFR, and PDC EFR are fixed-rate speech codec algorithms, and their rates are 12.20kbit/s, 7.40kbit/s, and 6.70kbit/s respectively. Since the algorithm principles of AMR, GSM EFR, TDMA EFR and PDC EFR all use the "arithmetic codebook excitation" (ACELP) method, and when designing the AMR decoding algorithm, the compatibility requirements are considered, so the AMR_12.20 mode and GSM EFR It is compatible; AMR_7.40 mode is compatible with TDMA EFR; AMR_6.70 mode is compatible with PDC EFR. But their silent frame (SID frame) in silent detection and discontinuous transmission mode is incompatible: the size of AMR_SID frame is 35bit and updated every 160ms, while the silent frame of GSM EFR is 244bit and updated every 480ms, The silent frame of TDMA EFR is 38bit and can be continuously updated. In addition, the DTX control of different voice codecs is also different. For example, TDMA EFR and PDC EFR only support uplink discontinuous transmission (DTX) in the original system, and do not support downlink DTX; and the DTX smoothing mechanism of AMR and other codecs Also different. Since the discontinuous transmission mode can effectively increase the radio channel capacity, save UE power, and save transmission resources, it is the default configuration item for UMTS devices. During codec negotiation, except for the specified codec rate, SID mode and NO DATA mode are set by default. Therefore, the incompatibility of the above-mentioned SID frames actually leads to the inability to achieve unified implementation of AMR and GSM EFR, TDMAEFR and PDC EFR.

The method of the present invention integrates the discontinuous transmission (DTX) and comfort noise generation functions of the GSMEFR, TDMA EFR and PDC EFR protocols on the basis of the core algorithm of the AMR voice codec, and realizes it uniformly through the control of the codec type and mode 7 voice codecs, the specific codec process is described as follows:

Fig. 1 is a schematic diagram of the encoding process used in the method of the present invention. As shown in Figure 1, the 13bit linear PCM (Pulse Code Modulation) data stream judges the current coding type according to the command set in step S101, and in the subsequent steps, the rate is set according to the coding type selected in step S101: if GSM EFR is selected Codec, then be set at the rate of AMR_12.20kbps in step S102-1; If select TDMA EFR, then be set at the rate of AMR_7.40kbps in step S102-2; If select PDC EFR, then set in step S102-3 Be determined as the rate of AMR_6.70; If select HR AMR, F AMR, UMTS AMR, UMTS AMR 2, then in step S102-4 to step S102-7 according to the rate that the rate control command is set the designated rate. In step S103: according to the speed of step S102 setting, carry out AMR speech encoding function, then judge in step S104 in the encoding process of step S103, judge according to calculating parameter and should output non-silent frame or silent frame, if judged to be non-silent, then Output voice compression stream in step S105; If it is judged to be silent, then judge current coding type in step S106: If judged to be GSMEFR, then in step S107-1, carry out the parameter of the algorithm calculation silent frame of GSM EFR, and in step S108- 1 output mute data according to the method of the discontinuous transmission of GSM EFR; If it is TDMA EFR, then in step S107-2, execute the algorithm of TDMA EFR to calculate the parameter of silent frame, in step S108-2 according to the method of discontinuous transmission of TDMA EFR Output mute data; if it is PDC EFR, then execute the algorithm of PDC EFR in step S107-3 to calculate the parameter of mute frame, and output mute data according to the method of the intermittent transmission of PDC EFR in step S108-3; If it is HR AMR, FR For AMR, UMTS AMR, and UMTS AMR 2, in step S107-4, the algorithm of UMTS AMR 2 is uniformly executed to calculate the parameters of the mute frame, and in step S108-4, the mute data is output according to the intermittent transmission method of UMTS AMR 2.

Decoding is the reverse process of encoding, only adding a wrong frame compensation and mute processing. Fig. 2 is a schematic diagram of the decoding flow used by the method of the present invention. As shown in Figure 1, in step S201: judge its bad frame indication (Bad Frame Indication-abbreviation BFI) mark from the received frame, check whether it is wrong frame; If step S201 judges to be normal, then program enters step S203 , if step S201 is judged to be wrong frame, then carry out wrong frame compensation and mute processing in step S202, judge its SID mark from the received frame in step S203; If judge in step S203 be silence, then judge current in step S204 Codec type: according to the codec type judged in step S204, call different mute processing procedures: if it is GSM EFR, then in step S205-1 and step S206-1 according to the intermittent transmission of GSM EFR and the SID parameter output mute data; If it is TDMA EFR, then in step S205-2 and step S206-2 according to the discontinuous transmission of TDMA EFR and the mute data of parameter output; Continuous transmission and SID parameter output mute data; If HR AMR, FRAMR, UMTS AMR, UMTS AMR 2, then in step S205-4 and step S206-4 according to the discontinuous transmission of UMTS AMR 2 and the output mute data of SID parameter; If step S203 is judged as non-silence, then judge current codec type in step S207; The codec type setting rate according to step S207 judgment: if be GSM EFR, then be set to the speed of AMR_12.20 in step S208-1; If If it is TDMA EFR, it is set to the rate of AMR_7.40 in step S208-2; if it is PDC EFR, it is set to the rate of AMR_6.70 in step S208-3; if it is HR AMR, FRAMR, UMTS AMR, UMTS AMR 2, then in step S208-4 according to the rate control command is set to specified rate; in step S209: according to the rate of S208 setting, in step S209 carry out the AMR speech decoding function; in step S210: output the linear PCM data of 13bit, so far, Describes the entire process of the codec program.

The present invention only needs to add comfort noise generation (SID) and discontinuous transmission (DTX) module and rate designation and control module of GSM EFR, TDMA EFR and PDCEFR on the basis of AMR algorithm, the realization workload of increase is very little, uses respectively The algorithm description program of the ANSIC language of AMR, GSM EFR, TDMA EFR, PDCEFR algorithm agreement is as the standard of workload, and Fig. 3 has described the amount of code that traditional realization method needs to develop, and Fig. 4 has described the required development of realization method of the present invention amount of code. Comparing the sum of the above two figures, the amount of codes to be implemented by the present invention is only 48% of that of the traditional implementation method, and the implementation cost (development time and expense) is greatly reduced.

In terms of storage space, the effect of the present invention is very obvious. Figure 5 refers to the application example on the TMS320C54X DSP, and gives the memory capacity required for the traditional implementation methods to realize all speech codecs. It can be seen that the method realized by the present invention saves 2.8 times in program space and 2.6 times in table space compared with the traditional method, and the effect is very obvious. According to the traditional implementation method, the space of 92K * 16bit is needed to realize the 1-way UMTS voice codec process, exceeding the on-chip memory capacity of most fixed-point DSPs; and the realization method of the present invention only needs the space of 35.4K * 16bit, which can be used in general fixed-point Realized on DSP. Therefore, the cost of hardware implementation is also greatly reduced.

Claims (3)

1, realize the method for multiple encoding and decoding speech in Universal Mobile Telecommunications System, it is characterized in that: described method comprises the steps:

During coding at first according to the current type of coding of command determination that is provided with;

Select rate mode according to type of coding;

Call unified adaptive multi-rate speech encoding and decoding (AMR) module then, carry out encoding operation;

Wherein, if determine it is quiet through silence detection, then call separately comfort noise coding (SID) and discontinuous transmission (DTX) module according to the respective coding type;

If right and wrong are quiet, then according to the speed output speech coding bit of determining;

Whether decoding is the inverse process of coding, at first judges its bad frame indication (BFI) sign from the frame that receives, check to judge whether to do wrong frame compensation according to bad frame indication (BFI) sign by wrong frame;

If judge it is wrong frame, then carry out wrong frame compensation and off beat and handle;

If judge it is normal, then judge whether it is quiet according to comfort noise coding (SID) sign, if quiet, then the corresponding code/decode type of basis calls separately discontinuous transmission and comfort noise decoder module;

If right and wrong are quiet, then speed is set according to corresponding code/decode type, carry out the AMR decode operation, unified at last output 13bit linear impulsive coded modulation (PCM) data.

2, method according to claim 1, it is characterized in that: described multiple be seven kinds, these seven kinds of encoding and decoding speechs are: gsm system enhanced full rate encoding and decoding (GSM EFR), tdma system enhanced full rate encoding and decoding (TDMA EFR), individual digital communication system enhanced full rate encoding and decoding (PDC EFR), half rate AMR encoding and decoding speech (HR AMR), full rate AMR encoding and decoding speech (FR AMR), Universal Mobile Telecommunications System AMR encoding and decoding speech (UMTS AMR), Universal Mobile Telecommunications System AMR encoding and decoding speech standard 2 (UMTSAMR 2).

3, method according to claim 1, it is characterized in that: described definite speed and according to corresponding code/decode type speed is set and is: GSM EFR selects AMR_12.20, TDMA EFR selects AMR_7.40, PDC EFR selects AMR_6.70, and HR AMR, FR AMR, UMTS AMR and UMTS AMR 2 then basis rate controlled order separately determine speed.

Images